JOURNAL OF GEOPHYSICAL RESEARCH, VOL. 99, NO.

D5, PAGES 10,641-10,652, MAY 20, 1994

Propagation effects on the lightning-generated electro-

magnetic fields for homogeneous and mixed sea-land paths
Vernon Cooray and Ye Ming
Instituteof High VoltageResearch,Universityof Uppsala,Sweden

Abstract. The influencesof the path of propagationon the shapeand amplitudeof the electric
fieldsandelectricfield derivativesgenerated by lightningreturnstrokeswereinvestigated. Results
arepresentedfor the caseswhere(1) boththe lightningreturnstrokeandthepoint of observation
are locatedover finitely conducting,homogeneous groundand (2) the lightningreturn strokeis
locatedover seaand the point of observationis locatedover land at somedistancefrom the sea-
land boundary.For propagationpathsover finitely conductinggroundthe electricfield derivative
can decreasesignificantlyin propagationdistancesas small as 1000 m. When the path of
propagationis partlyoverseaandpartlyoverland,thepropagation effectson the electricradiation
field derivativeare significantunlessthe width of the land portionof the propagationpath is less
than a few tens of meters.

1. Introduction km the attenuationof the peak is less than 5% for a ground

conductivityof 0.01 S/m and is less than 20% for a ground
Protection of structuresand electrical systemsfrom conductivityof 0.001 S/m. However,in that studyno attempt
lightningrequiresknowledgeconcerningthe characteristics was made to determine the propagation effects on the
of electromagneticfields generatedby lightning and the derivatives of the return stroke fields. Therefore we were
statisticaldistributionof lightningcurrentparameters.The interested in investigating the propagationeffects on the
statisticaldistributions
of lightningcurrentparameters
can be electric field derivatives of lightning-generatedelectro-
obtained by recording the currents in lightning flashes magnetic fields.
strikinghigh towers[Berger, 1967]. However,the presence To avoid the propagation effects caused by finitely
of the towermay distortthesedistributions to somedegree, conducting ground, many researchershave measured the
andthereis alwaysthe unresolved questionof whetherthese electromagneticfields from lightning using techniquesthat
distributionsare valid for lightningflashesstrikingplane minimize propagation effects. This was achieved, for
ground.On the otherhand,someof the information necessary example,by measuringelectromagnetic fields from lightning
for determiningthe characteristics of currentsin lightning flashesstrikingthe sea,so that the pathof propagation of the
return strokes is embedded in the lightning-generated electromagneticfields is over sea-water[Weidman and Krider,
electromagnetic fields. However, in propagatingfrom the 1980; Cooray, 1986; Willerr et al., 1988, 1990; Le Vine et
source to the measuring station, the characteristicsof the al., 1989]. Since seawater is a better conductor than soil,
electromagneticfields will change in various degrees ectromagneticfields propagatingover seawaterare subjected
dependingon the geometryand electricalcharacteristics
of to much less severe propagation effects compared with
the propagationpath. For example,in propagatingover electromagneticfields propagatingover land. Recently,Ming
finitely conductingground,the electromagneticfields will and Cooray[1993] analyzedthe propagation effectscausedby
lose their higher-frequency
components,and as a result,the a finitely conducting and rough sea on the lightning-
amplitude of the electromagneticfield decreasesand its generated electric fields. In some of the experimental
risetimeincreases
with increasingpropagation
distanceover investigationsmentioned above, the electromagneticfield
land [Le Vine et al., 1986; Urnanet al., 1976;Gardner, 1981; recordingsites were situatedseveral tens to several hundreds
Coorayand Lundquist,1983;Cooray, 1987].Thereforeany of metersinland [Cooray, 1986; Willerr et al., 1988, 1990;
attemptto derive return strokecurrentparametersfrom the Le Vine et al., 1989]. In such situations the path of
measuredfields without correctingfor thesepropagation propagationof electromagnetic fields generatedby lightning
effectsmay lead to significanterrors.Recently,Cooray flashesstrikingthe,sea is entirelyover seawaterexceptfor
[1987] analyzedthe propagationeffectson the risetimesand the last few tens or hundreds of meters. It would therefore be of
initialpea• of lightning
return
stroke
radiation
fieldsasthey interest to investigate the propagation effects on the
propagateover finitely conductingground.The resultsgiven electromagnetic fields causedby the presenceof this stripof
in that paper couldbe usedto estimatethe attenuationof the land in the path of propagation.
initial peakof the electricfield for distances
largerthanabout In this studywe calculatedthe propagationeffectson the
5 km. Furthermore, the resultsgivenin thatpapershowthat lightning-generated electric field derivativeswith special
the attenuationof the initial peak of the electric field is not attentionto distancesof propagationless than about 10 km.
very largefor distances smallerthan5 km. For example,at 5 Furthermore,resultsare also presentedfor the propagation
effectson the lightningreturnstrokefieldsas theypropagate
along a mixed sea-land path. These results allowed us to
Copyright1994by theAmericanGeophysical
Union. interpretthe experimentalobservations
obtainedby Le Vine
et al. [1989] and Willerr et al., [1988, 1990] where the
Papernumber931D03277. recording station was situatedseveral tens to severalhundreds
0148-0227/94/931D- 03277 $05.00 of metersfrom a sea-landboundary.
10,641
10,642 COORAY AND MING: PROPAGATION EFFECTS ON LIGHTNING RADIATION FIELDS

2. Theory isthespeed
of lightin vacuum,
R =(z 2+d2)mandsinO=z/R
For large distancesand small times the third term in the above
2.1. Basic Equations equation, i.e., the radiation field, is dominant [Urnan, 1987]
Let us assume that the lightning channel is straight and and the electric field is given by
vertical. The geometryunder considerationis shownin Figure
l a. The vertical electric field generatedat a horizontaldistance
d by a lightningreturn strokesituatedon perfectly conducting, 1 •osin
e,(d,t)=- 2Irœ'-•eR0o•i(z,
ce z'-R/c)
o•t dz (2)
flat, and smoothgroundis given by Uman [1987]

2.2. Expression for the Electric Radiation Field

e,(d,t)
=2zrœ
o RS i(z,x'-R/c)dzdz o
Under Different Propagation Conditions
Theoretical studies of electromagnetic wave propagation
H
over finitely conducting, homogeneousground have been
+I2-3sine
o 0 i(z, •'-R]c)dz given by Norton [1937, 1942], Wait [1962, 1964], Schlak
[1967], and King [1969]. The theory was extendedto the case
of a vertical dipole on the surfaceof an inhomogeneous
Earth,
i.e., a two-sectionmixed path, by Brernrner [1949] and Wait
isin
e0o•i(z,
o ceRz'-
R/c)
dz] o•t
(1)
[1956]. Up-to-date, complete descriptionsof the theory of
electromagnetic wave propagation over homogeneousand
inhomogeneousground were given by Wait [1986] and
where z is a coordinatedirectedalong the return strokechannel Maclean and Wu [1993]. In the presentstudy the theory, as
with the origin at ground level, H is the height of the return presented by King [1969], was used to describe the
stroke channel, i(z, t) is the current in the return stroke propagation over finitely conducting,homogeneousground.
channelat heightz, œ0isthe electricpermittivityof vacuum,c The resultsobtainedby King [1969] agree exactly with the
little known but more accurate work of Norton [1942]. The
Lighming theory as presentedby Wait [1956] was used to describe
channel propagation over a mixed sea-land path. Both of these
theorieshave been tested experimentallyby King and Maley
(a) [1966] and King et al. [1973a, b], and excellent agreement
has been found between the theory and the experimental
results.Now, let us incorporatetheseresultsinto the analysis
of propagation effects on lightning-generated
H
electromagnetic fields. Equation (2) can be written in
frequencydomain as
R

E,(d, jco)=-• 1 i sine

21reo
o ceR
0jco
l(z,jco).
exp(-jco R/c) dz (3)

where to is the angularfrequencyand I(z, jr o) is the Fourier

Lightning
channel transformationof i(z, t). That is,

(b)
I(z,jCO)
=-•-•I i(z,t)exp(-jco
t)dt (4)

dz
H When the groundis finitely conducting,the electricradiation
field at a horizontal distanced from the lightning channel,
E,.a(d,jco
), is givenby Wait[1956]andKing[1969]

A
.............................................
..............................................
E...(a. jco)= -• 1 isine
21reo 0jco
o ceRI(z,jco).
.............................................
..............................................
.........................
.........................
....................................
.............................................. W (jc0,z, d) exp(-jt0 R/c) dz (5)
..............................................................................
..............................................

Figure 1. Geometry relevant to the calculationspresentedin where W(jto, z,d)is the attenuationfunctioncorresponding
this paper. (a) Propagation of electromagnetic fields over to a vertical dipole situatedat a height z and a horizontal
homogeneous ground. (b) Propagation of electromagnetic distanced from the point of observation.This functiontakes
fields over a mixed sea-landpath. into account the propagation effects suffered by the
 COORAY AND MING: PROPAGATION EFFECTS ON LIGHTNING RADIATION FIELDS 10,643

electromagnetic field. When the ground is perfectly where

conducting, W(jro, z,d) = 1 and equation (5) reduces to

equation (3). Equation (5) can be transformedinto time F(jco,
d,Z)=1+j•-• exp(-p)erfc(-j•) (12)
domain, the result being
is the Sommerfeld's attenuation function and
Ht

ez.,(d,t)
=I I ea(•"z)w(t-
oo
•',z,d)d, (6)
4_••. T= sinO-Z• .
P= (1- T)2' sin
0+Z•'
where

j co
RZ2•
=Z/r/o;
2ccos2 0
ea(t,d)1osin
=2•rec22R0o•i(z,
t-
o•tR/c)dz (7)
and w(t, z, d)is the inverse Fourier transformation of
W (jro,z,d). That is, 0]
7o = -j co/c
w(t,z,d)
= I W(jco,
z,d)exp(jcot)
dco (8)
In the aboveequations,$to is the permeabiltyof free space,Z
Note that when the ground is perfectly conductingw(t, z, d) is thesurface
impedance
of themedium,Zn is thenormalized
reduces to a Dirac delta function or unit impulse function surfaceimpedanceof the medium,r/o is the impedanceof free
[Abramowitz and Stegun, 1972], and equation(6) reducesto space, and "erfc" is the complementary error function
equation(2). Note also that ed (t, z) is the electric radiation [Abrarnowitzand Stegun,1972]. In derivingequation(11), it
field generatedat a horizontal distanced by a dipole of length has been assumedthat [King, 1969]
dz at height z over perfectly conductingground.However, as
we will show below, in most of the casesof practical interest
the function W (jro,z,d) in equation(5) canbe replacedby the
Isin
o+z. 12
<<z
function correspondingto a dipole at ground level, i.e., Note that when z = 0,
W(jro, O,d). Under this approximation,equation (6) reduces
to
W (jco,O,d) = F(jco, d,Z) (14)
t

2.3.2. Propagation over a mixed sea-land path.

ez.,(d,t)
=I e,(d,•')w(t-•',O,d)d•
o
(9)
The geometryof the situationunder considerationis shown in
Figure lb. The lightningchannelis situatedover the sea at a
In thisequation,ez(d,t) is theradiationfield thatwouldbe distanced-dl from the coastline.The point of observationis
situated inland at a distance dl from the coastline. The
presentover perfectly conductingground,i.e.,
coastline is assumedto be straight, and the direction of
propagationof the electromagneticfields is assumedto be
1tr••osin
e'(d't)=- 2rc c:2
R0o•i(z,
•'-R/c)
t dz
o• (10) perpendicular to the coastline. An expression for the
attenuationfunctioncorresponding to a dipole at the ground
end of the lightning channelis given by Wait [1956, 1986]
and w(t, O, d) is the inverse Fourier transformation of as follows:

W (rio, O,d).

2.3. Expression for the Attenuation Function

w(jco,
o,a)
=F(jco,
a,Z)
-('Jcøct
) z,-z 2 rcc) %
Under Different Propagation Conditions
2.3.1. Propagation along plane and homo- I F(jCo,
d-x,Z)F(jro,x,Z•)
dx
geneous ground. The geometry under consideration is
similarto that in Figure l a. The groundis represented
by a
smooth,infinite half-spacewith conductivitycr and relative whereZ andZ 1 are the surfaceimpedances of seawaterand
dielectricconstantœ.The groundend of the lightningchannel land, respectively;F(jco,d,Z) and F(jco,d,Z•) are the
is locatedat A, and the point of observationis locatedat P. Sommerfeld's attenuation function (i.e., equation (12))
The attenuationfunctioncorresponding to a vertical dipole corresponding to seawaterand land, respectively.Note that
situatedat heightz from groundlevel, W(jco,z,d), is given when dl = 0, the propagationpath is completelyover the sea
by [King, 1969, equation (28)]. When the point of (i.e., a homogeneous medium)andthe aboveequationreduces
observation is at groundlevel, that equationreducesto to equation(14). In the derivationof this equationthe ocean
surface was assumedto be smooth, and as a result, any
propagation effects caused by the roughnessof the ocean
W(jco,
z,d)
=3[(1
+T)+(1-
T)F(jco,
d,Z)] (11) surface were neglected.However, our calculationsshow that
10,644 COORAYAND MING: PROPAGATIONEFFECTSON LIGHTNING RADIATION FIELDS

for propagation
distanceslessthanabout10 km overthe sea, and Rust, 1989] the modelpredictsa returnstrokevelocity
the propagationeffects on the electromagnetic fields are which decreaseswith height. The model-predictedaverage
mainly determinedby the strip of land in the propagation velocities over the first 300 m and 1000 m of the return stroke
path(seeFigure5a). Furthermore, calculations
by Ming and channelare 1.9 x 108m/s and 1.3 x 108 m/s, respectively.
Cooray [1993] showed that the hitherto available Thesevaluesare similar to thosemeasuredby Mach and Rust
experimentaldataconcerning the propagationeffectscaused [1989]andIdoneandOrville[1982].A detailed
comparison of
by an oceansurface[Weldmanand Krider, 1981] can be the model-predicted return strokeparameterswith those
explainedby assumingthat the wind velocity, which measured in experimental
investigations
wasmadeby Cooray
determines
theroughness
of the oceansurface,is about5 m/s. [1993].In thatpublication
it wasshownthatthereis a good
At this wind velocity, most of the propagationeffectsare agreementbetween the model predictionsand the
determined by the finitelyconducting
seawaterandnot by the experimentalresults.
roughness of the oceansurface.Thesetwo reasonsjustifythe
use of equation(15) which neglectsthe roughness of the
ocean surface. 4. Validity of the ApproximationsMade
in Deriving Equation (9)
3. ElectromagneticFields Generated In derivingequation(9), we made two assumptions: First,
we assumedthat the fields are pure radiation; second,we
by LightningReturn Strokes assumed that the errors introduced into the calculations are not
The propagationeffectsunderdifferentconditionscan be significant if W(jro,z,d) in equation (5) is replacedby
calculatedby solvingequation(9) if the temporalandspatial W(jro,O,d). To test the validity of the first assumption we
variation in the return stroke current, i.e., i(z, t), and the havecalculatedthe electricfield at differentdistances
by using
rerumstrokevelocityare known.The returnstrokemodelof equations(1) and (2). In thesecalculations,equations(1) and
Cooray [1993] is capableof predictingthe temporaland (2) were evaluatednumericallyby determiningthe integral
spatialvariationof thereturnstrokecurrentandreturnstroke over z. The procedureis identicalto dividingthe returnstroke
velocity.This rerumstrokemodelwasusedin performingthe channel into elementary dipoles and summing up the
calculationspresentedin this paper. The model-predicted contributionfrom each dipole (i.e., equation(7)), taking into
current waveform at three different heights are shown in accountthe time delay betweenthe contributionfrom different
Figure2 (sol/d,z = 0 m; shortdashed,
z = 100m; longdashed, dipoles.The currentin the dipolesat differentheights(i.e.,
z = 200 m). The time delaysbetweenthe beginningof the i(z, t)) was obtainedusing the return strokemodel [Cooray,
current waveforms are equal to the time taken by the return 1993]. Note also that the current in the return strokechannel
stroke front to reach the respectiveheights.This is the case is turnedon by the arrival of the return strokefront at each
since the current in the return stroke channel is turned on by height. Therefore the current in the dipolesabovethe return
the arrival of the return stroke front at that height. Note that stroke front is zero. For this reason the integral over z at a
the peak currentat groundlevel is 13 kA. Note alsothat the given time has to be determinedonly up to the heightof the
peak of the current waveform decreaseswith increasing return stroke front at that time. The position of the return
height.The peakcurrentderivativeof thecurrentwaveformat stroke front at a given time is also obtained by the return
groundlevel is 60 kA/}•s [Cooray, 1993]. Similar to the strokemodel. Comparisonof the resultsobtainedwith these
experimentalobservations[Idone and Orville, 1982;Mach equationsshowsthat the initial portion of the electric field
derivative,includingthe initial peak at distanceslarger than 1
16
km, can be completely described by equation (2). For
example, the electric field derivative calculated using
equations (1) (solidline) and(2) (dashedline) at 1 km from the
returnstrokechannelare depictedin Figure 3a. Similarly, our
•12
calculationsshow that at distanceslarger than 5 km the initial
portionof the electricfield, includingthe initial peak, can be
describedcompletelyby the radiationfield. For example,the
/
,,'
.-' electric fields calculatedusing equations(1) (solid line) and
(2) (dashedline) at 5 km from the return strokechannel are
..,, / depictedin Figure 3b. Note also that the resultsderivedby
King [1969] are valid only for the radiation(i.e., far field)
4 ! componentof the dipole fields. For this reason we have
restricted our calculations to distances at which the section of
the field, including the initial peak, can be describedby the
0 radiation field alone. This requiresd > 1 km for the electric
0 1 2 • ......... • ......... field derivative and d > 5 km for the electric field. Now, let us
considerthe secondassumption.To determinewhetherthis is
Figure 2. The model-predicted cu•ent waveform at •ree a reasonableassumption,we calculated the electric field
different heights (solid, z = 0 m; short-d•hed, z = 1• m; •d generatedby lightningreturnstrokesat differentdistancesby
1on•-d•hed, z = 2• m) • •e re• s•oke ch•el. •e pe• usingequations(6) and (9) for conductivities
in therangeof 4
cu•ent derivative of •e cu•ent wavefo• at •e •round level S/m to 0.001 S/m. In these calculations the relative dielectric
is 60 •/•s. The t•e delays between the bo•i•• of •e constantof the path of propagationwas assumedto be 81
cu•ent wavefo•s are equal to •e t•e t•en by •e re• when • = 4 S/m (i.e., seawater). For other conductivitiesit
sffoke front to reach •e respectiveheight. was assumedto be 5. The procedureof evaluatingequation(6)
 COORAY AND MING: PROPAGATION EFFECTS ON LIGHTNING RADIATION FIELDS 10,645

100 -

5oo0
]
4000. ,.
(a) (b)

"• 3000
'• 60

,•,2000
-,., 1000: ,.,

o 20

- 1000 0 'i11111 •i I • iili • II 1• Iilll • I "i • ilil ! 1111 •i •i

0.0 0.1 0.2 0.3 0.4 0.5 0.0 0.1 0.2 0.3 0.4 0.5

1600

(c) 5000
4000
(d)

•1200

,,.. 800 ,._ 2000

400 ./,,," ,1000

I I i I I I I I i I I II IIII III II I 1 il i ii i i i i I I i I i I I I I I - 1000

0.0 0.1 0.2. 0.3 0.4 0.5 0.0 0.1 0.2 0.3 0.4
T•,m• ('us)
Figure 3. (a) Electric field derivativegeneratedby a lightningreturn strokeat 1 km distancefrom the point
of strike over perfectlyconductingground.The solid curve was calculatedusingequation(1) and the dashed
curve was calculatedusing equation(2). (b) Electric field generatedby a lightning return stroke at 5-km
distancefrom the point of strike over perfectly conductingground.The solid curve was calculatedusing
equation(1) and the dashedcurvewas calculatedusingequation(2). (c) Electricfield derivativegeneratedby a
lightningreturn stroke at 1-km distancefrom the point of strike over finitely conductingground.The solid
curvewas calculatedusingequation(6) andthe dashedcurvewas calculatedusingequation(9) (ty= 0.001 S/m
andœ= 5). (d) Electricfield derivativegeneratedby a lightningreturnstrokeat 1-kmdistancefrom thepointof
strikeover finitely conductingground.The solid curvewas calculatedusingequation(6), and the dashedcurve
wascalculatedusingequation(9) (ty= 4 S/m andœ= 81).

is basicallysimilar to that used to evaluateequation(2). equations are similar. Indeed, our results show that the
However, in evaluatingequation(6) the contributionfrom assumptionis justified in calculating the attenuationof the
each dipole (i.e., equation (7)) is convoluted with the initial peakof the radiationfield (d >= 5 km) andits derivative
responseof the attenuationfunction (i.e., w(t, z, d)) before (d >= I km) for conductivitiesin the rangeof 0.001 S/m to 4
the contributionsfrom differentdipolesweresummedup. For S/m. Furthermore,since the approximationis valid for both
example,the electricfield derivativescalculatedat 1000m by sea paths (i.e., • = 4 S/m) and land paths (i.e., • = 0.01 to
using equations(6) (solid line) and (9) (dashedline) for the 0.001 S/m), the approximationis also valid for mixed sea-
conductivities0.001 S/m and 4 S/m are shownin Figures3c land paths [King et al., 1973b]. Furthermore, note that
and3d, respectively.Note thattheresultspredictedby the two equation(11) is derived under the approximationgiven by
 10,646 COORAY AND MING: PROPAGATION EFFECTS ON LIGHTNING RADIATION FIELDS

equation(13). Thereforein performingthe integrationover•z 1 km (curve 1), 5 km (curve 2), and 10 km (curve3) are shown
in equation (5), it is necessaryto make sure that this in Figure 4. The relative dielectric constantof the path of
condition remained satisfied.In the calculationspresentedin propagation is assumed to be 81 when ff = 4 S/m (i.e.,
this paper only the first microsecondof the fields are corresponds to seawater). For other conductivities it was
considered.During the first microsecondthe height of the assumedto be 5. First, note that for a given conductivityand
return stroke front is about 200 m. Since the current in the frequencythe amplitudeof the attenuationfunctiondecreases
dipoles above the return stroke front is zero, for all the with increasingdistances.Second,for a given distanceand
calculationspresentedin this paper, the integral in equation frequencythe amplitudeof the attenuationfunctiondecreases
(5) has to be performedonly up to a height of about200 m with decreasingconductivity.Third, for a given conductivity
(i.e., the maximum value of z in equation(5) is about200 m). and distance the attenuation function decreases with
Now, the closestdistancefrom the lightningchannelat which increasingfrequency.Considerthe curvescorrespondingto ty
we have performedour field calculationsis 1 km and as a result = 4 S/m. This conductivityis equal to that of seawater.These
the maximum value of sin 0 is = 0.2. Our calculations show curves show that in propagatingdistancesup to 10 km, no
that for this value of sin 0, equation(13) is satisfiedfor all the significant attenuation takes place for frequenciessmaller
conductivitiesand frequenciesof interest for the problem than about 10 MHz. On the other hand, propagationpaths
under considerationin this paper.However, as pointedout by over ground introducesignificant attenuationat frequencies
King et al. [1973a], the conditiongiven by equation(13) is higher than 500 kHz. The results show very clearly the
overly restrictive.They showedthat experimentaldata agree difficulty of obtainingthe spectrumof return strokeelectric
with
thetheoretical
data
even
when
Isin
0+Z,I approaches
1. fields at frequencies higher than about 1 MHz, even at
distancesas small as 1 km from the lighming return stroke.
5. Results and Discussion Now, let us considera sea-landpath of length d. Equation
(15) shows that the attenuation function correspondingto
The theory presented earlier shows that the effects of this path can be divided into two parts: The first part (i.e.,
propagation are introduced into the electromagneticfields F(jco,d,Z)) is the attenuationfunctioncorrespondingto a sea
throughthe attenuationfunction W(jco,O,d). Therefore, first path of distanced. The secondpart (i.e., the integral) takes
we will describehow the amplitudeof the attenuationfunction into accountthe presenceof the strip of land in the path of
varies with frequencyunder differentpropagationconditions. propagation.Figure5a showsthe amplitudeof the attenuation
Below we will show the effect of propagation on the function when d = 10.05 km and dl = 50 m. The results are
electromagneticfields generatedby lightningreturn strokes. given when the conductivityof the strip of land is 0.01 S/m
(short-dashedcurve) and 0.001 S/m (long-dashedcurve). The
5.1. Amplitude of the Attenuation Function amplitude of the attenuation function correspondingto a
Under Different Propagation Conditions completesea path of length 10.05 km is also given in the
The amplitudes of W(joo,O,d) in decibel (i.e., figure for comparativepurposes(solid curve). Comparisonof
2OlogolW(jo,O,d) path with thesecurvesshowsthat attenuationof the high frequenciesis
I) for a homogeneous
conductivitiesof 4 S/m (solid curve), 0.01 S/m (short-dashed mainly causedby the presenceof the strip of land in the path
curve), and 0.001 S/m (long-dashedcurve) and for distancesof of propagation.Figure 5b showsthe resultswhen the length
of the path of propagationis 10.2 km and dl = 200 m. Figures
5c and 5d show the resultswhen d = 30.05 km, dl = 50 m and
m lO when d = 30.2 km, dl = 200 m, respectively.Thesediagrams
123 provide important information concerningthe attenuationof
the frequenciesin electromagnetic
fields when they propagate

-10
-=,., '.......
-......... alongdifferentsea-landpaths.They alsoshowthe importance
of keeping the electromagnetic field propagation path
. x '. ',, 123 completely over the sea in attempts to investigate the
-20 frequencyspectrumof electromagneticfields generatedby
lightning return strokes.
-30
123
-40
5.2. Propagation Effects on the Electromagnetic
Fields Generated by Lightning Return Strokes
-50 Now, let us considerthe effectsof propagation-induced
attenuation
of high frequencies
on the electromagnetic
fields
E -60 generatedby returnstrokesoverfinitelyconducting
ground.
"• 10 • 10 ? First,we will considerthepropagation
effectsintroduced
by
Fr e q•zer•c3t (Hz)
homogeneous paths.Belowwe will describe
thepropagation
Figure 4. Amplitude of the attenuationfunction in decibel effectscaused
by mixedsea-land
paths.
(i.e., 2ologolW(jo,o,a)l),
for propagation
overhomo- 5.2.1. Propagation effects caused by finitely
geneousground.Solidcurve,(or= 4 S/m andœ= 81); curve1, d conducting, homogeneous ground.
Let us consider
= 1 km; curve2, d = 5 km; andcurve3, d =10 km. Short-dashed the effectsof propagation
on lightning-generated
electric
curve,(or= 0.01 S/m andœ= 5); curve1, d= 1 km; curve2, d = 5 fields as they propagateover finitely conducting,
km; andcurve3 d =10 km. Long-dashed curve,(ty= 0.001S/m homogeneousEarth. The propagation effects on the
andœ= 5); curve1, d= 1 km; curve2, d = 5 km; andcurve3, d derivativeof the electricfield in propagating1 km, 2 km, 5
-10
km, and10 km whenthe groundconductivity is 0.01 S/m are
 COORAYAND MING: PROPAGATIONEFFECTSON LIGHTNING RADIATION FIELDS 10,647

o
o

-2

-5

-10

-15

-20
10 • 10 7
•vr e qwer•c l/ (Hz) (b) 10' ' rreq'u,
encv(Hz)10•

-4
-8

-8 -12

-16

-12
-20

-16 -24
10 • 10 7
(c) Freqizer•c!t (Hz) (d) 10' rreq•encv
(mz)10•
Figure
5.Amplitude oftheattenuation
function,
indecibel
(i.e.,201Og•o[W(jco,
O,d)[),
forpropagation
over
amixedsea-landpath.(a)d = 10.05
km,dl= 50m,cr= 0.01S/m(short-dashed
curve)
or0.001S/m(long-
dashed
curve).(b)d = 10.2km,dl= 200m,ty= 0.01S/m(short-dashed
curve)or0.001S/m(long-dashed
curve).
(c)d= 30.05km,dl= 50m,cr= 0.01S/m(short-dashed
curve)or0.001S/m(long-dashed
curve).
(d)d
= 30.2km,dl= 200m,cr= 0.01S/m(short-dashed
curve)
or0.001S/m(long-dashed
curve).
In each
case the
solidcurvecorresponds
to theattenuation
function
fora complete
seapathof length
d. Notethatcristhe
conductivity
of thelandportionof thepropagation
path.

shownin Figure6. In the calculations

the relativedielectric Krider [1981],the propagation
pathwasentirelyover
constantof the ground was kept constantat 5. For seawater, andin theexperiments
carriedoutby Willettet al.
comparativepurposesthe electricfield derivativecalculated [1990],theposition
of themeasuring
stationwassuchthat
overperfectlyconducting
ground is alsoshownin thefigure thepathof propagation
wasentirely
overseawater
exceptfor
(curve1). Sincetheradiation
fielddecreases
inversely with thelast50 m. Letusconsider
theeffects
of propagation
over
distance
in the absence
of propagation
effects,theresultsin an oceansurfaceon the spectrumof the returnstrokes.The
Figure6 arenormalized
to 100km by multiplying
the effectsof transmission
overa mixedsea-land
pathwill be
calculatedfieldamplitudes
by d/105whered is thedistanceto consideredbelow.Figure8 shows thespectrum of theelectric
the returnstrokein meters.Thereforeall the waveformsin field, normalized to 100 km, for different distancesof
Figure6 wouldhavethesameshape andamplitude if there propagation
overa smooth sea(curve2, d = 10km;curve3, d
wereno propagation
effects.Theresultsfor cr= 0.001S/mare = 30 km; andcurve4, d = 50 km). In thiscalculation
we have
shownin Figure7. Note thattheelectricfield derivativecan assumedthat cr = 4 S/m andœ= 81. The spectrum was
decrease
byabout
70%inpropagating
1000m andby90%in calculatedby inverseFourier transformationof the electric
propagating10 km overfinitelyconducting
ground
withty= fields,and0 dB corresponds
to 1 V/m/Hz.Forcomparative
0.001 S/m.
purposes the spectrum
calculated
overperfectlyconducting
Recently,Weidmanand Krider [1981]andWillettet al. ground(curve1) and the experimentaldataof Weidmanand
[1990]extended
thereturnstroke
spectrum to frequencies
Krider[1981](dashed curve)arealsoshown in thisdiagram.
higher
thanabout20MHz.In these
studiesthespectrum
was Sincethemeasurements of Weidrnan andKrider[1981]were
calculated
fromtheelectromagnetic
fieldspropagated
overan fromlightningreturnstrokesat distances
of 40 - 50 km, the
oceansurface.
In theexperiment
conducted
by Weidrnan
and agreement between the calculated spectrum and the
 COORAY AND MING: PROPAGATION EFFECTS ON LIGHTNING RADIATION FIELDS 10,649

land portion betweenthe sea-landboundaryand the point of

observation is less than a few hundred meters. However, the
propagationeffects on the electric radiation field derivative
are significant unlessthe width of the land portion is less
than a few tens of meters.
Now, let us consider the experimentallymeasuredreturn
stroke spectrumof Willett et al. [ 1990]. As mentionedearlier,
(a : / 1 in this experimentthe location of the measuringstation was
'+ i i such that the path of propagationwas entirely over seawater
exceptfor the last 50 m. Figure 11 showsthe spectrumof the
electric field calculatedfor different values of (d-dl), i.e., 10
1. km (curve 2), 20 km (curve 3), and 30 km (curve 4), while
keeping dl constant at 50 m. In this calculation the
conductivity and the relative dielectric constant of the soil
O. was assumedto be 0.005 S/m and 5, respectively.Note that
o.o o.• o.z o.s o. 4 o.s
the differencein the spectralamplitudesat 20 MHz when (d-dl)
Time (microsec. )
is equalto 10 km and when (d-dl) is equalto 30 km is about4
dB. This is in good agreementwith the resultspublishedby
Willett et al. [1990]. For comparativepurposesthe spectrum
calculatedover perfectly conductingground(curve 1) and the
experimental data of Willett at al. [1990] (dashedcurve) are
also shown in this diagram. Note that the spectrum, as
published by Willett et al. [1990], was given in terms of
spectraldensity.Therefore as suggestedin their paper, the
dataof Willett et al. [1990] had to be decreasedby 3 dB before
they could be comparedwith ours. Since the measurements of
Willett et al. [1990] were from lightning return strokesat
distancesless than 35 km and the effect of the roughnessof
the oceansurfaceis to decreasethe spectralamplitudeat 20
MHz for 30-km propagation
by about3 dB [Ming and Cooray,
1993], the agreementbetweenthe calculatedand the measured
O.
valuescan be consideredgood.
0.0 0.1 0.œ 0.$ 0.4 0.5 The results presented above are also of interest in the
Time (microsec. ) interpretationof the electric field parametersmeasuredby
Willett et al. [1988] for rocket-triggeredlightning. In their
experimentthe locationof the triggeringsite and recording
stationwere such that the path of propagationwas entirely
over brackishwater except for the last few hundredmeters.
The resultspresentedaboveshow that in propagatingalong
this path, the derivativeof the electricfield can decreaseby
about 10% - 50% dependingon the soil conductivity.For
example, if the strip of the land is 100 m and the soil
-o /' • conductivity is 0.005 S/m (which is typical for soil in
•2 / 2 Florida),the derivativecan decreaseby about20%. Of course,
ß i (c)
it is possiblethat the strip of land in the experimentis not
homogeneous, as assumed in the above theoretical
calculation.For example,the strip of land may consistof an
upperlayer of soil and a highly conductingwater table below
it. However, for soil with a conductivity of 0.005 S/m the
O. i i i I i i
o.o o.• o.z o.• skindepthat frequenciesabout 10 MHz is about2 m. Therefore
Time (microsec. if the depth of the upper layer exceeds about 2 m, the
propagationeffectsfor frequencieshigher than about 10 MHz
Figure 9. (a) Effects of propagatingalong a mixed sea-land
would not differ significantly from those calculated for
path on the electric fields generated by lightning return
homogeneousground.
strokes. Curve 1, dl = 50 m; curve 2, dl = 100 m; curve 3, dl =
200 m; curve4, dl = 500 m. In this calculationthe lengthof
5.3. Effects of Propagation on the Remote
the path of propagationover the sea (i.e., d-dl) was kept
constantat 5 km, and the conductivityof the soil was assumed Sensing of Lightning Return Stroke
Current Parameters
to be 0.01 S/m. The dashed curve is the electric field calculated
over perfectly conducting ground. All the results are Recent theoreticaland experimentalinvestigationsshowed
normalized to a common distanceof 100 kin. (b) Same as in that the lightningcurrent and lightningcurrentderivativecan
Figure 9a except that the soil conductivityis 0.005 S/m. (c) be obtainedfrom the measuredfields using transmissionline
Same as in Figure 9a except that the soil conductivity is model (TLM) equations[Urnanand McLain, 1970;Willett et
0.001 $/m. al., 1988; Rakov et al., 1992; Cooray, 1993]. According to
 10,650 COORAY AND MING: PROPAGATION EFFECTS ON LIGHTNING RADIATION FIELDS

-1 oo
50.

40. -•20
Ul
(a)
30.
•.,-14-0
,,.

Z -•6o
•' •.0. o
• -i80
10. •o

O.
-200 ß
,,.,-.2
\ ,-,-.$
'x,•"4
I
-I0........ 0.0,,. O. I 0.2 0.3 0.4
,....... -220
0.5
10 •
Freq•ertc!l (Hz)
10 ?

Time (microsec. ) Figure 11. Effects of propagatingalong a mixed sea-land

path on the spectrumof the electric fields generatedby

•$0..
!
ß

lightningreturn strokes.In the calculationthe lengthof the

landpathandthe conductivity of the soil werekeptconstant
at
•,
•
o
40. /',,,,,
, ß (b) 50 m and 0.005 S/m respectively. Lengthsof the seapaths
(i.e., d-dl): curve 2, 10 km; curve 3, 20 km; and curve 4, 30
-•_ 30.0
km. Curve 1 is the spectrumof the electricfield calculatedover
perfectly conductingground.For comparativepurposesthe
•x2o.
0 spectral data, as measuredby Willett et al. [1991] are also
shown in the figure (dashed curve). All the results are
normalized to a common distance of 100 km.

•,1o. 0.
theTLM equations thepeakof theelectric radiation field,Et,'
thepeakderivative of theradiationfield,(dE/dOt,,thepeak
-•0. current,
It,, andthepeakcurrentderivative,(di/dt)t,,arerelated
o.o o. t rime
o.• o.•
(mi crosec. )
o.• o.s by

50. Ep=2•o
zrd
vlp (16)

• 4o. , (C) (aœ1at),,=

Sto
21rd v(ai/at),, (17)

30ø whered is distanceto themeasuringstationandv is thereturn

strokevelocity. In estimatingthe currentparametersfrom
•' •,0. these equations,it is necessaryto know the effects of
propagation on the electric radiation fields and electric
radiationfield derivatives.
In severalpublications,
Cooray
• tO. and Lundquist[1983]andCooray[1986] analyzedhow the
amplitudeof the radiationfield decreases in propagating
O.
differentdistancesover land.The informationpresented
in
thosepapers can be used to estimate the errors introducedin
estimatingthereturnstrokepeakcurrents
by using equation
o.o o. I o.ß o.3 o.4 o.a (16) together
withtheradiationfieldsmeasured
overfinitely
Time(micr osec.) conducting
ground.
Recently,
Willett
etal.[1988]
obtained
a
Figure 10. (a) Effects of propagating along a mixed sea- formulathat relatesthe measuredfield peaksto the return
land path on the electric field derivatives generated by strokepeakcurrent.The formulawasobtained by analyzing
lightning return strokes.Curve 1, dl = 50 m; curve2, dl = 100 the simultaneouslymeasured lightningcurrentsandradiation
m; curve 3, dl = 200 m; curve 4, dl = 500 m. In this calculation fields.The propagationpathof thefieldsusedby Willettet al.
the length of the path of propagationover the sea (i.e., d-dl) [1988]wascompletely oversaltwaterexceptfor thelastfew
was kept constantat 5 km, and the conductivityof the soil hundredmeterswhichwas over ground.The propagation
was assumed to be 0.01 S/m. The dashed curve is the electric effectsthat can be expectedfrom sucha path are shownin
field derivative calculatedover perfectly conductingground. Figures9a through9c. The formulathereforeaccounts,
at least
All the results are normalized to a common distance of 100 partly,for attenuation
because
themeasuredfieldsarealready
km. (b)Same as in Figure 10a except that the soil attenuated.However, the attenuationsuffered by these
conductivityis 0.005 S/m. (c) Same as in Figure 10a except waveformsis smallcomparedwith the attenuationthatcan be
that the soil conductivityis 0.001 S/m. expected
overfinitelyconducting
groundfroma propagation
 COORAY AND MING: PROPAGATIONEFFECTSON LIGHTNING RADIATION FIELDS 10,651

positive return strokes and their submicrosecondstructure, J.

patha fewtensof kilometers long. Forexample, if thelength
Geophys.Res.,91, 7907-7911, 1986.
of thestripof landis 100m andtheconductivity is 0.005S/m Cooray,V., Effectsof propagationon the rerumstrokeradiationfields,
(typicalforFloridasoil)thenthedecrease in theamplitude of Radio Sci., 22(5), 757-768, 1987.
thepeakdueto propagation effectsis lessthan2% (Figure Cooray, V., A model for subsequentrerum strokes,J. Electrost.,30,
9b). On the otherhand,in propagating 50 km overfinitely 343-354, 1993.
conducting ground of conductivity
0.005S/mtheamplitude of Cooray,V., and Lundquist,S., Effectsof propagation on the risetimes
the radiation field will decrease more than 10% [Cooray, and the initial peaks of radiationfields from return strokes,Radio
1987].The attenuation
of theradiationfieldwill be higherfor Sci., 18(5), 405-415, 1983.
longer distancesand smaller conductivities. Therefore Gardner, R. L., Effects of propagationpath on lightning induced
transientfidds, Radio Sci., 16, 337-384, 1981.
significant
errorsmayresultif the formulais usedtogether
Idone, V. P., and R. E. Orville, Lightning rerum strokevelocitiesin the
with the radiationfields that have propagatedseveraltensof
ThunderstormResearchInternationalProgramORIP), J. Geophys.
kilometersover finitely conducting
ground.The
propagation Res., 87, 4903-4915, 1982.
effects are more critical in estimating the peak current Jayaratne,K. C. P., and V. Cooray, Effects of propagationacrossa
derivative from the electric field derivatives. For example, coastline on the electromagnetic
fields from lightningrerum strokes,
resultspresented
in thispapershowthatthederivative
of the paperpresentedat the 20th InternationalConferenceon Lightning
electricfield can decrease
by about70% in propagating
1 km Protection (ICLP), Swiss Electrotech. Assoc., Interlaken,
overfinitelyconducting
groundof conductivity
0.001 S/m. Switzerland, 1990.
Therefore if the derivative of the electricfield measuredfrom a King, R. J., Electromagnetic wave propagation over a constant
lightning
returnstrokeat 1 km oversuch ground
is usedin impedanceplane,RadioSci.,4, 255-268, 1969.
King, R. J., and S. W. Maley, Model experimentson propagationof
equation(17) to estimatethe returnstrokepeak current
derivative,the estimationwill be 3 timesas smallas the peak groundwaves
acrossan abruptboundaryat obliqueincidence,
Radio
Sci., 1,111-115, 1966.
current derivative in the return stroke. These results show the
King, R. J., S. H. Cho, D. J. Jaggard,and V. Sokolov,Height-gain
necessity
of conducting
experiments
designed
to estimate experimentaldata for groundwave propagation,1, Homogeneous
(di/dt)p
from
themeasured
fields
under
maritime
conditions
so paths,RadioSci.,8, 7-15, 1973a.
thatthe effectsof propagation
areminimal. King, R. J., S. H. Cho, andD. J. Jaggard,Height-gainexperimentaldata
for groundwave propagation,2, Heterogeneous paths,Radio Sci., 8,
6. Conclusions 17-22, 1973b.
Le Vine, D. M., L. Gesell,andM. Kao, Radiationfrom lightningretum
In this article we have described how the path of strokes over a finitely conducting earth, J. Geophys. Res., 91,
propagationinfluencesthe shapeand amplitudeof the electric 11,897-11,908, 1986.
fields and electric field derivatives generatedby lightning Le Vine, D. M., J. C. Willett, and J. C. Bailey, Comparisonof fast
electricfield changesfrom subsequent return strokesof naturaland
return strokes.Results are presentedfor the caseswhere (1)
triggeredlightning,J. Geophys.Res.,94, 13,259-13,265,1989.
both the lightningreturn stroke and the point of observation Mach, D. M., and W. D. Rust, Photoelectricreturnstrokevelocity and
are locatedover fruitely conducting,homogeneous groundand peakcurrentestimatesin natureandtriggeredlightning,J. Geophys.
(2) where the lightning return stroke is located over the sea Res. 94, 13,237-13,247, 1989.
and the point of observationis locatedover land at a distance Maclean, T. S. M., and Z. Wu, Radiowavepropagationover ground,
dl (50 m - 500 m ) from the sea-landboundary.The results ChapmanandHall, London,1993.
show that for distances as small as 1 km the propagation Ming, Y., and V. Cooray,Propagationeffectscausedby a roughocean
effects on the electric field derivative cannot be neglected. surfaceon the electromagnetic fields generatedby lightning return
When the pathof propagationis partly over the seaandpartly strokes,in press,RadioSci., 1994.
over land, the propagationeffects on the the peak of the Norton,K. A., Propagationof radiowavesoverthe surfaceof the earth
radiation field can be neglected if the width of the land andtheupperatmosphere, I1, Proc.IEEE, 25, 1203-1236,1937.
Norton, K. A., The polarizationof downcomingionosphericradio
betweenthe coastalboundaryand the point of observationis
waves,FCC Rep. 60047, Nat. Bur.of Stand.,Boulder,Colo., 1942.
less than about 100 m. However, the propagationeffects on Rakov, V. L., R. Thottappillil, and M. A. Uman, On the empirical
the electric radiation field derivative are significantunlessthe formulaof Willerr et al. relatinglightningreturnstrokepeak current
width of the land portion in the path of propagationis less andpeakelectricfield,J. Geophys. Res.,97, 11527-11533,1992.
than a few tens of meters. Schlak, G. A.,An investigationof electromagnetic wave propagation
over an inhomogeneousground,Ph.D. thesis, Univ. of Colorado,
Boulder, 1967.
Acknowledgments. The researchwork presentedin this paper is Uman, M. A., The lightning discharge,Academic,San Diego, Calif.,
supported
by a grant(G-EG/GU 1448-306)from the SwedishNatural 1987.
Science Research Council. Thanks are due to Victor Scuka for the
Uman, M. A., and D. K. McLain, Lightning return strokecurrentfrom
research facilities placed at our disposal. We also thank two magnetic and radiation field measurements,J. Geophys.Res., 75,
anonymousrefereeswho providedcommentswhichhelpedto improve 5143-5147, 1970.
the manuscript. Uman, M. A., C. E. Swanberg,J. A. Tiller, Y. T. Lin, and E. P. Krider,
Effectsof 200 km propagationon lightningreturn strokeelectric
fields, Radio Sci., 11,985-990, 1976.
References
Wait, J. R., Mixed pathgroundwavepropagation,1, Shortdistances, J.
Abramowitz,M., and Stegun,M., Handbookof mathematicalfunctions, Res.Natl. Bur. Stand.,57(4), 1-15, 1956.
Dover, Mineola, New York, 1972. Wait, J. R., The propagation of electromagnetic
wavesalongthe Earth's
Berger, K., Novel observationson lightning discharges:Results of surface,in ElectromagneticWaves,editedby R. E. Langer,pp. 243-
researchon Mount San Salvatore, J. Franklin Inst., 238, 478-525, 290, Universityof WisconsinPress,Madison,1962.
1967. Wait, J. R., Electromagneticsurface waves, in Advanced in Radio
Bremmer, H., Terrestrial radio waves, Elsevier, New York, 1949. Research,editedby J. A. Saxton,vol. 1, pp. 157-217,Academic,San
Cooray,V., A novelmethodto identifythe radiationfieldsproducedby Diego, Calif., 1964.
10,652 COORAY AND MING: PROPAGATIONEFFECTSON LIGHTNING RADIATION FIELDS

Wait,J.R.,Propagation
effects
fordectromagnetic
pulsetransmission, line model of electromagneticradiationfrom triggeredlightning
Proc.IEEE, 74, 1173-1181,1986. retumstrokes,
J. Geophys.
Res.,93, 3867-3878,1988.
Weldman,C. D., andE. P. Krider,Submicrosecond structureof the Willett, J. C., J. C. Bailey,C. Leteinturier,andE. P. Krider,Lightning
retumstroke waveforms, Geophys.Res.Lett.,7, 955-958,1980. electromagnetic radiationfieldspectrain theintervalfrom0.2 to 20
Weidman,C. D., andE. P. Krider,Theamplitude of lightning MHz, J. Geophys.
spectra Res.,95, 20,367-20,387, 1990.
radiation
fieldsin theinterval
from2 to 20 MHz,Radio$ci.,21(6),
964-970, 1986.
Weidman,
C. D., E. P. Krider, amplitude V. CoorayandY. Ming,Instituteof HighVoltageResearch,
andM. A. Uman,Lightning 752 28
spectra
intheinterval
from100kHzto20MHz,Geophys.
Res.
Lett., Uppsala,Sweden.
8, 931-934, 1981.
A. Eybert- (ReceivedJanuary18, 1993;revisedOctober20, 1993;
Willett,J. C., V. P. Idone,R. E. Orville,C. Leteinturier,
Berard, andE. P. Krider,Anexperimental testof thetransmissionacceptedNovember16 1993.)